EP1331621A1 - Method for monitoring the environment of a vehicle - Google Patents

Abstract

The vehicle environment supervision system calculates at a supervision system measurement point an assembly of possible future positions of the point. The calculation takes account of the uncertainty of the measurement position. A measurement of the speed of the point furnished is provided.

Description

The present invention relates to motor vehicles, and more specifically monitoring the environment of such vehicles, to determine the dangerousness of a potential obstacle. Depending on the results obtained, it is possible to trigger safety equipment, such as cushions inflatable or automatic braking, or to warn the driver of the vehicle.

US-A-5,314,037 describes a collision avoidance system for a vehicle; the vehicle is equipped with one or more laser radars, which detect the position, direction of movement, speed and size of obstacles. It is also planned to use a steering angle sensor or gyroscope to provide information on the direction variations of the vehicle equipped with the system. The system calculates an estimated collision instant, by the ratio between the distance of the vehicle to obstacle and the relative speed of vehicle to obstacle. This estimated instant is compared to a minimum admissible time, the calculation of which involves a multivariate function. The nature of this function is not specified. In depending on the nature of the comparison, different alarms are implemented.

US-A-6,085,151 describes a predictive collision sensor system. This document suggests methods for analyzing obstacles and their behavior. The targets are identified in the cluster analysis detection system in English) and are followed by Kalman filtering in a Cartesian coordinate system. We calculate, for the tracks obtained at the output of the Kalman filter, predictions for different times, corresponding to the deployment times of the different vehicle safety equipment. The threat posed by each lead is evaluated from the point closest to the trajectory. A probability of threat is assigned to each object according to the accuracy of the track, its condition and the track quality. These different parameters take into account the uncertainty associated with Kalman filtering with the predictive method used, but are not representative of the errors generated by the detection system.

There is therefore a need for a reliable, simple and inexpensive system which allows to process the results provided by a supervision system of the environment of a vehicle.

Consequently, the invention proposes, in one embodiment, a method for processing the results provided by an environmental supervision system of a vehicle includes the calculation, for a point whose position is measured by the environmental supervision system, a set of possible positions of the point at a future time, the calculation taking into account the uncertainty of the measurement of the position of the point by the supervision system.

According to a variant, the supervision system provides, for a point, a measurement of the velocity of the point, and in that the calculation takes into account the uncertainty of the measurement of the point speed by the supervision system.

According to another variant, the calculation takes into account the uncertainty on the acceleration of the point.

According to yet another variant, the acceleration of the point is bounded by a value predetermined maximum.

• le calcul d'une position moyenne du point à l'instant futur à partir de la position et de la vitesse mesurées par le système de supervision; et
• le calcul d'un ensemble de positions autour de cette position moyenne à partir de l'incertitude sur la position mesurée et sur la vitesse mesurée.

According to yet another variant, the calculation comprises:

• the calculation of an average position of the point at the future instant from the position and the speed measured by the supervision system; and
• the calculation of a set of positions around this average position from the uncertainty on the measured position and on the measured speed.

It can also be foreseen that the method further comprises a calculation step of a probability of collision with the vehicle at the future instant, by summing the probabilities of presence of the point at the different positions of the intersecting set the vehicle.

• la division de l'ensemble à des positions possibles de chaque point en au moins deux sous-ensembles;
• le tracé de segments de droites reliant les sous-ensembles correspondants des ensembles des deux points;
• le calcul d'une probabilité de collision de l'objet et du véhicule en sommant, pour un des deux points, les probabilités de présence du point dans les différents sous-ensembles reliés par des segments de droites intersectant le véhicule.

According to a variant, the method comprises for an object two points of which are detected by the environmental supervision system:

• dividing the set at possible positions of each point into at least two subsets;
• drawing line segments connecting the corresponding subsets of the sets of the two points;
• the calculation of a probability of collision of the object and the vehicle by summing, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle.

• la division de l'ensemble des positions possibles de chaque point en au moins deux sous-ensembles ;
• le tracé de segments de droites reliant les sous-ensembles correspondants des ensembles de deux points adjacents;
• le calcul, pour chaque paire de poins adjacent, de la somme, pour un des deux points, les probabilités de présence du point dans les différents sous-ensembles reliés par des segments de droites intersectant le véhicule; et
• le calcul d'une probabilité de collision de l'objet et du véhicule comme la valeur maximale de la probabilité calculée pour chaque paire de points adjacents.

According to another variant, the method comprises, for an object of which three or more points are detected by the environment supervision system:

• dividing the set of possible positions of each point into at least two subsets;
• drawing line segments connecting the corresponding subsets of the sets of two adjacent points;
• the calculation, for each pair of adjacent points, of the sum, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle; and
• calculating a probability of collision of the object and the vehicle as the maximum value of the probability calculated for each pair of adjacent points.

• le calcul d'une probabilité de collision avec le véhicule selon de tels procédés de traitement;
• le déclenchement de l'équipement en fonction de la probabilité calculée.

The invention also relates to a method for triggering safety equipment in a vehicle, comprising:

• calculating a probability of collision with the vehicle according to such processing methods;
• the triggering of the equipment according to the calculated probability.

• une routine de réception de la position et le cas échéant de la vitesse d'un point fournis par le système de supervision;
• une routine de détermination d'une incertitude sur la position et le cas échéant d'une incertitude sur la vitesse du point;
• une routine de calcul, pour un point, d'un ensemble de positions possibles du point à un instant futur, à partir de la position reçue et de l'incertitude déterminée sur la position.

The subject of the invention is also a program for processing the results supplied by a system for monitoring the environment of a vehicle, comprising:

• a routine for receiving the position and if necessary the speed of a point provided by the supervision system;
• a routine for determining an uncertainty on the position and, where appropriate, an uncertainty on the speed of the point;
• a routine for calculating, for a point, a set of possible positions of the point at a future time, from the received position and the uncertainty determined on the position.

According to a variant, the calculation routine is a routine for calculating the set possible positions from the received position and the uncertainty determined on position and uncertainty on the acceleration of the point.

• une routine de calcul d'une probabilité de collision avec le véhicule à l'instant futur, par sommation des probabilités de présence du point aux différentes positions de l'ensemble intersectant le véhicule.

According to another variant, the program also includes:

• a routine for calculating a probability of collision with the vehicle at the future instant, by summing the probabilities of presence of the point at the different positions of the assembly intersecting the vehicle.

• une routine de division de l'ensemble des positions possibles des deux points en au moins deux sous-ensembles;
• une routine de tracé de segments de droites reliant les sous-ensembles correspondants des ensembles des deux points;
• une routine de calcul d'une probabilité de collision de l'objet et du véhicule par sommation, pour un des deux points, les probabilités de présence du point dans les différents sous-ensembles reliés par des segments de droites intersectant le véhicule.

According to yet another variant, the reception routine is adapted to receive an indication that an object is detected by two points; and in that it further comprises:

• a routine for dividing the set of possible positions of the two points into at least two subsets;
• a routine of drawing line segments connecting the corresponding subsets of the sets of the two points;
• a routine for calculating a probability of collision of the object and the vehicle by summation, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle.

• une routine de division de l'ensemble des positions possibles de chaque point en au moins deux sous-ensembles;
• une routine de tracé de segments de droites reliant les sous-ensembles correspondants des ensembles de deux points adjacents;
• une routine de calcul, pour chaque paire de poins adjacent, de la somme, pour un des deux points, les probabilités de présence du point dans les différents sous-ensembles reliés par des segments de droites intersectant le véhicule; et
• une routine de calcul d'une probabilité de collision de l'objet et du véhicule comme la valeur maximale de la probabilité calculée pour chaque paire de points adjacents.

According to yet another variant, the reception routine is adapted to receive an indication that an object is detected by at least three points; and in that it further comprises:

• a routine for dividing the set of possible positions of each point into at least two subsets;
• a routine of drawing line segments connecting the corresponding subsets of the sets of two adjacent points;
• a routine for calculating, for each pair of adjacent points, the sum, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle; and
• a routine for calculating a probability of collision of the object and of the vehicle as the maximum value of the probability calculated for each pair of adjacent points.

The invention further relates to a system for triggering vehicle security, including an on-board calculation system with memory containing a program as described above.

• figures 1 et 2, une représentation schématique d'un véhicule et d'un objet détecté par un système de supervision de l'environnement,
• figure 3, une représentation schématique des positions possibles de l'objet dans le futur, selon un mode de réalisation de l'invention;
• figure 4, des exemples pour le calcul d'une probabilité de collision;
• figure 5, un graphe d'une probabilité de collision en fonction du temps;
• figure 6, une représentation d'un objet détecté par plusieurs points du système de supervision de l'environnement;
• figure 7, des exemples pour le calcul d'une probabilité de collision avec un objet détecté par plusieurs points du système de supervision de l'environnement;
• figure 8, un ordinogramme des étapes du procédé de l'invention.

Other characteristics and advantages of the invention will appear on reading the following description of embodiments of the invention, given by way of example and with reference to the appended drawings which show:

• FIGS. 1 and 2, a schematic representation of a vehicle and an object detected by an environmental monitoring system,
• FIG. 3, a schematic representation of the possible positions of the object in the future, according to an embodiment of the invention;
• Figure 4, examples for calculating a probability of collision;
• FIG. 5, a graph of a probability of collision as a function of time;
• FIG. 6, a representation of an object detected by several points of the environment supervision system;
• FIG. 7, examples for calculating a probability of collision with an object detected by several points of the environment supervision system;
• FIG. 8, a flow chart of the steps of the method of the invention.

The invention proposes, for a vehicle provided with a system for supervising the environment, to take into account, in the analysis of the results provided by the system, the uncertainty that affects these results. This uncertainty may have source of errors in measuring the position of the object, the relative speed of the object in relation to the vehicle or other information used. She can also have for source the future behavior of the object or the vehicle. Taking into account these uncertainties allow a calculation of the probability of collision, for a detected object by one or more points of the environmental supervision system.

In the following description, the invention is described with reference to a vehicle automobile; it is more generally likely to apply for any system supervision of the environment, even if it equips a fixed device. Such supervision system is known per se, and is not described in more detail here: it can in particular to be a system based on radars, of the type described in the documents of the state of the art mentioned above. We use in the following, except mention on the contrary, an orthonormal reference, linked to the vehicle; the direction x is the direction of the vehicle speed relative to the ground, the direction being perpendicular to it.

Figure 1 shows a schematic representation of a vehicle and an object one point of which is detected by an environmental monitoring system. We have shown in Figure 2 a vehicle 2, provided with the environmental monitoring system. Reference 4 designates a point detected by the supervision system of the environment; this system typically identifies such a point by its coordinates in the benchmark.

Position measurement in the environmental monitoring system is affected by uncertainty; the position of point 4 is therefore only known with a uncertainty, which can in particular be a function in a laser system, of the width of the beam, of the fineness of the scanning by the laser beam and of the distance at which lies the point. The position of the detected point is therefore not necessarily exactly that shown in the figure by reference 4.

Figure 1 therefore shows, around point 4, a set 6 of positions possible from the detected point; in the example, this set has the shape of a rectangle, whose width and length are representative of the precision with which point is detected. This form of the set 6 of the possible positions corresponds to an embodiment in which point 4 is identified by coordinates (x, y) in a Cartesian coordinate system; the set 6 can then be constituted by the points whose abscissa is in the interval [x - δx; x + δx] and whose ordinate is in the interval [y - δy; y + δy], with δx and δy the respective details on the measured positions. The set 6 could have a different shape, depending the environmental supervision system used; for example, if the system of supervision uses radial scanning by a laser beam, the set 6 could have the shape of a crown sector; the angle of this sector would correspond to uncertainty about the angle of the laser, while the thickness of the crown would correspond to uncertainty about distance measurement. Set 6 could also be defined, at instead of considering possible ranges of coordinates, by a collection of points; these points can be distributed more or less uniformly around point 4 measured according to the supervision system used.

In the example proposed in Figure 1, we consider a set 6 of positions possible from the object corresponding to point 4 detected. It is also possible to take take into account the precision of the measurement. So, for a system of environmental supervision, it may be more likely that the object is located point (x, y) detected rather than a point with coordinates (x - δx; y - δy), corresponding to a corner of rectangle 6; this can be taken into account by adding to the set 6 a density of probability of presence of the object; this may correspond for example to a Gaussian centered on the measured point.

Thus, the set 6 makes it possible to take into account the uncertainties on the position the point detected by the environmental monitoring system; these uncertainties depend on the supervision system; they can be stored, or calculated in function of points. The position of the object corresponding to the detected point is no longer only provided by point 4 detected, but is provided by all 6 of possible positions.

The environmental monitoring system can also provide a measure of the speed of a detected obstacle; it is often a speed relative to the vehicle in the case of an on-board detection system; taking into account the speed of the vehicle, it is also possible to provide an absolute speed of the obstacle in the terrestrial reference frame. Like the position, this speed is affected by an uncertainty: Figure 1 shows two vectors 8 and 10, the vector 8 represents the speed v ₀ measured while vector 10 is an example of another possible speed vector taking into account the uncertainty in the speed measurement. As in the case of position, we can adopt various representations to define a set of possible speeds; one can in particular consider ranges of possible values of the coordinates of the speed vector in a coordinate system, a collection of possible speed vectors. One can also store the possible uncertainties, or even calculate them.

The figure also shows, by the dotted lines 12, all the positions possible of the object detected in the future, taking into account the uncertainties on the position and on speed; the figure still shows two extreme positions A and B, which are the extreme positions of the assembly 12. Depending on the measurement errors, the real object could have a position in the set 12 which more or less approaches the points A or B. If the object is in the vicinity of point A, there will be no collision, while if it is near point B, there will be collision. Because detection uncertainties of position and / or speed, it is not possible to a priori in the example of Figure 1 to know if the detected object will or not strike the vehicle.

In a first embodiment, the invention therefore proposes to take into account consideration, in an environmental supervision system, of the uncertainties of measured. Depending on the supervision system and, where applicable, the results of the measurement, we can take into consideration only the uncertainties on the position, only uncertainty about speed, or both.

• t : temps d'extrapolation ;
• Posx(t) : la composante suivant x de la position relative d'un point de l'objet à l'instant t ;
• Posy(t) : la composante suivant y de la position relative d'un point de l'objet à l'instant t ;
• Posx₀ : la composante suivant x de la position relative d'un point de l'objet à l'instant t₀ ;
• Posy₀ : la composante suivant y de la position relative d'un point de l'objet à l'instant t₀ ;
• Vitrelx₀ : la composante suivant x de la vitesse relative de l'objet à l'instant t₀ ;
• Vitrely₀ : la composante suivant y de la vitesse relative de l'objet à l'instant t₀ ; (t0 est l'instant de l'acquisition des données)

This can be represented by an average trajectory, affected by an uncertainty. The average trajectory is then given in a Cartesian coordinate system linked to the vehicle by the equations: Posx (t) = Posx 0 + Vitrelx 0 .t Posy (t) = Posy 0 + Vitrely 0 .t with the following notations, t ₀ being the instant of data acquisition

• t: extrapolation time;
• Posx (t): the component along x of the relative position of a point on the object at time t;
• Posy (t): the component y of the relative position of a point on the object at time t;
• Posx ₀ : the component along x of the relative position of a point of the object at time t ₀ ;
• Posy ₀ : the component following y of the relative position of a point of the object at time t ₀ ;
• Vitrelx ₀ : the component along x of the relative speed of the object at time t ₀ ;
• Vitrely ₀ : the component y of the relative speed of the object at time t ₀ ; (t0 is the moment of data acquisition)

• Incx(t) : Incertitude sur x de position à l'instant t ;
• Incy(t) : Incertitude sur y de position à l'instant t ;
• Errrposx : Erreur sur la position en x pouvant être commise par le système de supervision;
• Errrposy : Erreur sur la position en y pouvant être commise par le système de supervision;
• Errvitrx : Erreur sur la vitesse en x pouvant être commise par le système de supervision;
• Errvitry : Erreur sur la vitesse en y pouvant être commise par le système de supervision.

The uncertainty around the average trajectory can be given by the following equations: Incx (t) = Errposx 0 + Errvitrx.t Incy (t) = Errposy 0 + Errvitry.t with the following notations:

• Incx (t): Uncertainty on position x at time t;
• Incy (t): Uncertainty on y of position at time t;
• Errrposx: Error on the x position that can be committed by the supervision system;
• Errrposy: Error on the position, which can be committed by the supervision system;
• Errvitrx: Error on the speed in x that can be committed by the supervision system;
• Errvitry: Error on the speed which could be committed by the supervision system.

Such modeling provides a set of possible trajectories, with a look of the kind shown at 12 in the figure. The line 14 represents the average trajectory of the obstacle.

Figure 2 also shows a schematic representation of a vehicle and a object detected by an environmental monitoring system; we represented, as in Figure 1, vehicle 2, point 4 detected, all 6 positions possible and two speed vectors 8 and 10 possible. In the embodiment of Figure 2, we take into account not only the uncertainties on the measurement of position or speed by the environmental monitoring system but also uncertainties about future behavior; in other words, in the extrapolation of the relative trajectory of the detected obstacle - or of all the relative trajectories possible of the obstacle - we take into account possible actions of the driver of the vehicle or object speed changes, i.e. events likely to occur between the time of detection and the possible time of collision with the vehicle.

In other words, this embodiment proposes to take into account not only average behavior, but also specific events and suddenly likely to change the future evolution of objects.

This uncertainty about the behavior of the object or the vehicle can be modeled by the fact that an acceleration - longitudinal or transverse - can occur at any time, for an indefinite period. However, in the case of automotive detection systems, it is accepted that acceleration is limited by adhesion, whether for braking or steering. It is therefore possible to model future behavior by a limit on the relative longitudinal or transverse acceleration. This limit can be fixed at a given value - for example 10 ms ^-2 for a motor vehicle and an obstacle, or be updated regularly by a system capable of measuring parameters such as grip on the ground.

• Incaccx : la composante suivant la direction x de l'incertitude sur l'évolution de la vitesse, modélisée par l'accélération maximale relative de l'objet suivant la direction x; et;
• Incaccy : la composante suivant la direction y de l'incertitude sur l'évolution de la vitesse, modélisée par l'accélération maximale relative de l'objet suivant la direction y.

For such a model, there is an uncertainty around the average trajectory given by the following equations: Incx (t) = Errposx 0 + Errvitrx.t + Incaccx.t 2 / 2 Incy (t) = Errposy 0 + Errvitry.t + Incaccy.t 2 / 2 with the following notations:

• Incaccx: the component along the direction x of the uncertainty on the evolution of the speed, modeled by the maximum relative acceleration of the object along the direction x; and;
• Incaccy: the component along the y direction of the uncertainty about the evolution of the speed, modeled by the maximum relative acceleration of the object along the y direction.

The relative maximum acceleration of the object in the direction x is assumed negative - in a braking strategy to avoid the object; maximum acceleration relative of the object in the y direction can be positive or negative.

We again obtain a set of possible trajectories, around the average trajectory 14; this set of possible positions is shown in the figure under reference 16. The lateral limits of all the possible trajectories are parables.

Figure 3 shows another possible representation of the possible positions of the object in the future; the figure shows like figures 1 and 2 the vehicle 2, the detected point 4 and the set 6 of the possible positions at the instant t ₀ . It also shows all the possible positions at different times t ₁ to t ₇ , which are referenced 20, 22, 24, 26, 28, 30 and 32 in the figure. These different sets can be obtained by the same equations as those proposed with reference to Figure 2; we simply consider not the possible trajectories, but the possible points at a given instant.

The embodiments described so far take into account, in the predicting the trajectories of an obstacle in relation to a vehicle fitted with a system of environmental supervision, the uncertainties linked to the measurement or future behavior of the vehicle and the obstacle.

It thus becomes possible to define a probability of collision between the vehicle and the obstacle; this probability can be provided, for a set of trajectories, by assigning a probability to each trajectory, and summing the probabilities of trajectories meeting the vehicle. A simpler calculation of the probability of collision is now described with reference to Figure 4. This figure shows the vehicle 2, as well as three sets 36, 38 and 40 of possible positions of a obstacle; it can be possible positions of the same obstacle to different moments, or positions of different obstacles.

• la surface de l'intersection entre le véhicule et l'ensemble des positions possibles, et
• la surface de l'ensemble des positions possibles de l'obstacle.

In this embodiment, the probability of collision with an obstacle is calculated by considering the proportion of all the possible positions occupied by the vehicle. In other words, we calculate the probability of collision by considering the relationship between

• the area of the intersection between the vehicle and all of the possible positions, and
• the surface of all the possible positions of the obstacle.

In the example of set 36 in the figure, the probability of collision is zero, at time t corresponding to this set. In fact, whatever the position of the object in the assembly 36, it does not come into contact with the vehicle. Generally 40 of the figure, the vehicle covers all the possible positions; the probability collision is therefore 100%.

In the example of the assembly 38 in FIG. 4, part of the assembly of positions of the object is in the space occupied by the vehicle, it cannot be said with certainty whether or not there will be a collision; we then calculate the probability as being the portion 39 of the assembly 38 included in the space occupied by the vehicle, which in the example is around 40%.

In the example above, the probability for set 38 is obtained by a simple surface ratio; this solution is suitable when the probability of presence of the point in the set of possible positions is uniformly distributed. In the event that the measurement error of the supervision system is not uniformly distributed around the measured value, we can take this distribution in the calculation of the probability of collision. The calculation can then be a surface integral of the probability of presence in the whole, limited to the surface intersected by the vehicle. In either case, calculating the probability amounts to summing the probabilities of presence of the point at the different positions of all the possible positions intersecting the vehicle. In the case of a uniform probability distribution, the calculation amounts to a surface calculation; for a point cloud, the calculation is a discrete sum; for non-uniform distribution of probability, the calculation is a surface integral.

For an obstacle, the calculation can be carried out for different instants, after the instant of the measurement; in the example of figure 3, one could calculate for each of the represented sets, ie for each of the times t ₁ to t ₇ the probability of a collision, with the method described with reference to figure 4. Thus, it is possible to know the evolution of the probability of collision as a function of time; this probability can be represented in the form of a graph of the kind of that of FIG. 5. This one shows time on the abscissa, and on the ordinates the probability of collision.

Such a calculation can be done continuously or discretely, with a time step larger or smaller depending on the needs of the protection system using information. This information is then made available to the vehicle protection, for ordering safety equipment or for generating driver alerts.

In the above description, we have considered the example of an obstacle detected by a single point. However, some supervision systems are capable of describing by more than one point each detected object, or certain objects. These systems provide a finer representation of the outline of the object. Figure 6 shows a representation of an object detected by several points of the environmental monitoring system; object 44 - a post in the example - is detected several times by the environmental supervision system, which detects three points M ₁ , M ₂ and M ₃ , at the positions referenced 46, 48 and 50 at the instant t ₀ . The environmental monitoring system can also provide an indication that the detected points belong to the same object.

As described above, each point can be assigned a set of possible positions, as shown by 52, 54 and 56 in Figure 6.

The trajectory prediction can also be applied to each of these points, in the same way as described above: from a trajectory prediction which takes account of the uncertainties and measurement errors, we determine, for each instant t to come, the set of possible positions of each point of the object. FIG. 6 therefore shows the sets 58, 60 and 62 of the possible positions of the three points M ₁ , M ₂ and M ₃ , at an instant t ₁ . FIG. 1 shows a calculation of the kind described in FIG. 1, for reasons of simplicity of the figure; we could also carry out a calculation of the kind described with reference to FIG. 2. We have also shown in the figure the average trajectory and the envelope of the possible trajectories between the instants t ₀ and t ₁ .

It is advantageous, when calculating the probability of collision, to take into account that the points belong to the same object, and that part of the object is present between two successive positions. In other words, we consider that the object detected is not punctual, but has a certain volume, and we correlate the collision probabilities between the different points of the object.

For this, in one embodiment, it is to divide each set of possible positions into subsets, and to connect the corresponding subsets. This is shown in FIG. 6, in the example of the assemblies 58 and 60 for the points M ₁ and M ₂ . Each set 58 and 60 is divided into 4 subsets, with an identical probability of presence in the simplest case. Because the detected points belong to the same object - which is assumed to be undeformable - if a detected point is in one of the subsets, then the other point is in the same subset. For example, if at time t ₁ the point M ₁ is in the upper left subset (in the direction of the figure) of the set 58, then the point M ₂ is also in the subset upper left (in the direction of the figure) of the assembly 60.

• en déterminant ceux des portions ou segments qui sont intersectés par le véhicule, et
• en sommant la probabilité de présence d'un point dans chacun des sous-ensembles ainsi intersectés.

We then connect the corresponding sub-assemblies, as shown in Figure 6; this defines a portion or segment connecting the two subsets. The probability of collision with a vehicle can be assessed

• by determining which of the portions or segments are intersected by the vehicle, and
• by summing the probability of the presence of a point in each of the subsets thus intersected.

Figure 7 shows an example of calculating such a probability; we represented in the figure two examples of the case of an object detected by two points of the system of environmental supervision; sets 64 and 66 are sets of places possible of these two points in a first example, while the sets 68 and 70 are the sets of possible locations of the two points in a second example; in each case, these sets were divided into twenty-five subsets. For the simplicity of the explanation, we assume a uniform distribution of the probability of presence in these subsets, so the probability of presence of a point in a subset is 1/25. The figure also shows the vehicle 2. In the first example, the vehicle intersects all segments or portions connecting the subsets of sets 64 and 66; the probability of collision with the object detected by the colon is 100%; on the other hand in the second example, the vehicle intersects 22 of the 25 segments or portions connecting the sub-assemblies; the probability of collision with the object detected by the two points is 22/25.

The method of FIG. 7 thus makes it possible to calculate a probability of collision for an object detected by two or more points. For an object detected by two points, we proceed to the calculation represented in figure 7. For an object detected by more than two points, we can repeat the calculation of figure 7, by pair of adjacent points of the object. Adjacent points are points of the same object that follow one another in a scan of the object by the supervision system. Thus, in the example of FIG. 6, we would first proceed to the calculation of the intersected segments connecting the subsets of sets 58 and 60, then to the calculation of the intersected segments connecting the sub-sets of sets 60 and 62. The probability of collision with the obstacle is in this case the highest probability obtained for a pair of points of the obstacle.

This solution can be used regardless of the detection means used work and the extrapolation method used. As explained with reference to the figure 6, the calculation of this probability is advantageously made for each instant which one wishes to know the probability of shock. Frequency and accuracy - directly function of the number n of subsets - of the calculation is to be optimized by depending on the application using the information.

FIG. 8 is a flow chart of the steps of the method of the invention, in the example of an environmental monitoring system that allows detection of the same object by several points. The flow chart relates to the treatment applied to an object, the processing can be repeated for all the detected objects. In a first step 72, the position data and if necessary the speed data are received relative of the different points, as a function of time. From this data, in a step 74, the sets of possible positions are calculated for each time step different points detected; we separate these sets into subsets, and we calculates line segments connecting two sets of positions of two points adjacent. In step 76, we test whether the line segments intersect the contour of the vehicle, and we calculate the probability of each subset. Then, in step 78, the probability of shock is calculated for each pair of adjacent points. At the stage 80, we calculate the probability of impact with the object, considering the maximum value of the probability for the different pairs of adjacent points. The probability of shock can then be transmitted, in step 82, to be used in order to control safety equipment or alert the driver.

From a material point of view, the invention can be implemented from environmental monitoring systems known per se; just add it, in a processor or any other means of calculation, uncertainty values for allow the calculation of the sets of positions described above. The other stages of calculation are implemented in the calculation means or means, according to the explanations provided above.

The calculation system can be implemented using a dedicated processor or another means of calculation. It implements a treatment program for information provided by the sensors, which is stored in the associated memory; this program presents a routine for receiving the position and, if applicable, the speed of a point provided by the supervision system; this is suitable for receive the positions and speeds of the detected points; if applicable, it is also adapted to receive from the environmental monitoring system an indication that certain points detected correspond to the same object.

The program also presents a routine for determining an uncertainty on the position and if necessary an uncertainty on the speed of the points. We can use tables stored in memory, or function calculation algorithms including the position of the points detected, the number of times they were detected, etc.

The program finally presents a routine for calculating, for a point, a set of possible positions of the point at a future time, starting from the position received and uncertainty determined about the position; where appropriate, the speed information and uncertainties about speed, as well as limit values allowable acceleration. The calculation can be carried out according to one of the modes of realization described above.

• une routine de division de l'ensemble des positions possibles des deux points en au moins deux sous-ensembles;
• une routine de tracé de segments de droites reliant les sous-ensembles correspondants des ensembles des deux points; et
• une routine de calcul d'une probabilité de collision de l'objet et du véhicule par sommation, pour un des deux points, les probabilités de présence du point dans les différents sous-ensembles reliés par des segments de droites intersectant le véhicule.

The program can also present one or more routines for calculating the probability of collision; in the case of an object detected by a single point, the routine can calculate the probability of collision as explained with reference to FIG. 4, by summing the probabilities of presence of the point at the different positions of the assembly intersecting the vehicle. In the case of an object represented by two points, the routines can proceed as explained with reference to Figures 6 and 7. We can then provide:

• a routine for dividing the set of possible positions of the two points into at least two subsets;
• a routine of drawing line segments connecting the corresponding subsets of the sets of the two points; and
• a routine for calculating a probability of collision of the object and the vehicle by summing, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle.

• une routine de division de l'ensemble des positions possibles de chaque point en au moins deux sous-ensembles;
• une routine de tracé de segments de droites reliant les sous-ensembles correspondants des ensembles de deux points adjacents;
• une routine de calcul, pour chaque paire de poins adjacent, de la somme, pour un des deux points, les probabilités de présence du point dans les différents sous-ensembles reliés par des segments de droites intersectant le véhicule; et
• une routine de calcul d'une probabilité de collision de l'objet et du véhicule comme la valeur maximale de la probabilité calculée pour chaque paire de points adjacents.

In the case of an object identified by more than three points of the supervision system, one proceeds by pair of adjacent points, and the routines can present:

• a routine for dividing the set of possible positions of each point into at least two subsets;
• a routine of drawing line segments connecting the corresponding subsets of the sets of two adjacent points;
• a routine for calculating, for each pair of adjacent points, the sum, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle; and
• a routine for calculating a probability of collision of the object and of the vehicle as the maximum value of the probability calculated for each pair of adjacent points.

Routine here means a set of instructions capable of controlling the operation of a processor to perform a function. The program is stored in the memory of an on-board calculation system; it is executed in a known manner in itself by loading into the RAM of the calculation system.

The frequency of information processing may be set by the frequency of the environmental supervision system, and is preferably greater than 40Hz, for sufficient reactivity of the system.

The proposed solution takes into account the errors and uncertainties of the environmental supervision. It is therefore not necessary to use a system extremely complex and very precise to apply this solution, and to have a reliable probability of collision information. Such information may to manage the triggering of passenger compartment protection means. So, we will avoid false alarms because we will know if the collision will really take place or no. On the other hand, one can also generate an alarm very early for systems reversible that can be triggered even if the risk of collision is less than 100%; this may be the case for automatic braking systems, or for vehicle driver alert systems.

Another advantage of the solution is to be valid whatever the object dynamics, on the front or side of the vehicle, for fixed objects or mobile.

Of course, the present invention is not limited to the examples and methods of realization described and represented, but it is susceptible of numerous variants accessible to those skilled in the art. We can notably modify the numerical values proposed in the different examples. The order of the different stages of the Figure 8 can be modified: for example, we can calculate the probabilities attached to each subset before calculating the line segments connecting the subassemblies. We can also use other methods of calculating uncertainties than those mentioned above, depending on the supervision systems considered; the point sets can also have shapes or representations different from those proposed in the figures.

Claims (15)

A method for monitoring the environment of a vehicle (2) comprising: the measurement by a system of environmental supervision of the position of at least one point and the calculation, for a point whose position (4) is measured by the environmental supervision system, of a set (6) of possible positions of the point at a future time, the calculation taking into account the uncertainty of measurement of the position of the point by the supervision system. The method of claim 1, characterized in that the supervision system provides, for a point, a measurement of the speed of the point, and in that the calculation takes into account the uncertainty of the measurement of the speed of the point by the supervision system. The method of claim 1 or 2, characterized in that the calculation takes into account the uncertainty on the acceleration of the point. The method of claim 3, characterized in that the acceleration of the point is limited by a predetermined maximum value. The method of one of claims 2 to 4, characterized in that the calculation comprises: the calculation of an average position of the point at the future instant from the position and the speed measured by the supervision system; and the calculation of a set of positions around this average position from the uncertainty on the measured position and on the measured speed. The method of one of claims 1 to 5, characterized in that it further comprises a step of calculating a probability of collision with the vehicle at the future instant, by summing the probabilities of presence of the point at the different positions of the assembly intersecting the vehicle. The method of one of claims 1 to 5, characterized in that it includes, for an object, two points of which are detected by the environmental monitoring system: dividing the set (58, 60) at possible positions of each point into at least two subsets; drawing line segments connecting the corresponding subsets of the sets of the two points; calculating a probability of collision of the object and the vehicle (2) by summing, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle. The method of claim 7, characterized in that it comprises, for an object of which three or more points are detected by the environment monitoring system: dividing the set (58, 60, 62) of the possible positions of each point into at least two subsets; drawing line segments connecting the corresponding subsets of the sets of two adjacent points; the calculation, for each pair of adjacent points, of the sum, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle; and calculating a probability of collision of the object and the vehicle as the maximum value of the probability calculated for each pair of adjacent points. A method of triggering safety equipment in a vehicle, comprising: calculating a probability of collision with the vehicle according to the method of claim 6, 7 or 8; the triggering of the equipment according to the calculated probability. A program for processing the results provided by a system for monitoring the environment of a vehicle, comprising: a routine for receiving the position and if necessary the speed of a point provided by the supervision system; a routine for determining an uncertainty on the position and, where appropriate, an uncertainty on the speed of the point; a routine for calculating, for a point, a set of possible positions of the point at a future time, from the received position and the uncertainty determined on the position. The program of claim 10, characterized in that the calculation routine is a routine for calculating the set of possible positions from the received position and the determined uncertainty on the position and an uncertainty on the point acceleration. The program of claim 10 or 11, characterized in that it further comprises: a routine for calculating a probability of collision with the vehicle at the future instant, by summing the probabilities of presence of the point at the different positions of the assembly intersecting the vehicle. The program of claim 10 or 11, characterized in that the reception routine is adapted to receive an indication that an object is detected by two points; and in that it further comprises: a routine for dividing the set of possible positions of the two points into at least two subsets; a routine of drawing line segments connecting the corresponding subsets of the sets of the two points; a routine for calculating a probability of collision of the object and the vehicle by summation, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle. The program of claim 10 or 11, characterized in that the reception routine is adapted to receive an indication that an object is detected by at least three points; and in that it further comprises: a routine for dividing the set of possible positions of each point into at least two subsets; a routine of drawing line segments connecting the corresponding subsets of the sets of two adjacent points; a routine for calculating, for each pair of adjacent points, the sum, for one of the two points, the probabilities of presence of the point in the different subsets connected by segments of lines intersecting the vehicle; and a routine for calculating a probability of collision of the object and of the vehicle as the maximum value of the probability calculated for each pair of adjacent points. A system for triggering vehicle safety equipment, comprising an on-board calculation system with a memory containing a program according to one of claims 10 to 14.

Images